Northern Goshawk Broadcast Surveys: Hawk Response Variables and Survey Cost

Studies in Avian Biology No. 16:24-30, 1994.

NORTHERN GOSHAWK BROADCAST SURVEYS: HAWK RESPONSE VARIABLES AND SURVEY COST

SUZANNEM. JOY, RICHARD T. REYNOLDS, AND DOUGLAS G. LESLIE

Abstract. We examined responsesof Northern Goshawks(Accipter gentilis) to taped broadcastcalls of conspecificsin tree-harvestareas and around alternategoshawk nests on Kaibab National Forest, Arizona, in 199 1 and 1992. Forest areastotaling 476 km2 were systematicallysurveyed for goshawks. Ninety responsesby adult and juvenile goshawkswere elicited and 15 active nestswere located. NO differencein responserates between sexeswas detected.Adult males, however, tended to approach the broadcasterwithout vocalizing,whereas adult femalesapproached while vocalizing.Our success in finding active nests after getting responsesfrom females was greater than after male responses. Goshawksresponded more often to broadcastsduring the nestling(2.0 responses/100stations) than fledgling(1.0 responses/100stations) period. During the fledglingperiod, adults were more likely to respondto broadcaststhan juveniles. Total costsassociated with our surveysfor goshawkswere $58- 82 per km2 and $4.15-5.80 per broadcaststation, dependingon salaries. Key Words: Accipitergent&; broadcastsurveys; Kaibab Plateau;Northern Goshawk;vocalization.

In the southwestern United States, Northern (Rassmussen1941). Forests on the Kaibab Plateau Goshawks (Accipiter gentilis) use ponderosa pine consistof a band of pinyon-juniper (1103 km*) at el- (Pinus ponderosa), mixed coniferous-deciduous evations between 1830-2075 m, ponderosapine forests(624 km*) between 2075-2500 m elevation, and forests, spruce-fir, and pinyon-juniper (Pinus ed- mixed-coniferforests (Abies concolor, Picea engelman- ulis-Juniperus spp.) woodlands. These forests are nii, Populustremuloides, Pseudotsuga menziesii) (605 subject to structural and compositional changes km*) above 2500 m (Rasmussen1941). A complete due to plant growth and succession, and to var- descriptionof the studyarea was provided in Reynolds ious natural (e.g., fire) and anthropogenic (e.g., et al. (this volume). tree harvest, grazing) disturbances that may af- BROADCASTS fect goshawk reproductive success (Reynolds 1983, 1989; Cracker-Bedford 1990, Reynolds et We used a modification of the Kennedy and Stah- al. 1992). In 1982 the Southwestern Region of lecker (unpubl. data) broadcastprotocol to survey for goshawks.Goshawk vocalizations were broadcast from the USDA Forest Service designated the gos- stationsestablished on parallel transects,and their re- hawk as a sensitive species.Reynolds et al. (1992) sponseswere followed by searchesfor activenests (con- suggestedhabitat-management strategies for both taining eggsor young).Responses were classifiedas (1) goshawks and their prey in ponderosa pine and vocalizationonly, (2) sightingonly, or (3) vocalization mixed-species forests. The first step in the man- and sighting.We used cassetteplayers (199 1 surveys) agement of habitat for goshawks is locating their and portable, long-range callers (1992 surveys) to nests. broadcastgoshawk vocalizations. Cassettes were played Broadcasting of raptor vocalizations has been at a volume that produceda soundaudible to the hu- used to locate goshawk nests (Kimmel and Yah- man ear at a minimum of 150 m from the source. Rain, windsexceeding 20 km/h, and occasionalvehicle ner 1990, Mosher et al. 1990, Kennedy and Stah- traffic resulted in delay or termination of the survey lecker 1993). Broadcasts elicit vocal and/or vi- effort becauseof the interferencewith responsedetec- sual cues of raptors, which can then be followed tion. with nest searches. During the 199 l-l 992 breed- Surveyswere only conductedduring the nestlingand ing seasons we conducted broadcast surveys for fledgling periods when goshawk response rates to goshawks on 12 proposed tree-harvest areas on broadcastsare highest(Kennedy and Stahlecker1993). the North Kaibab Ranger District (NERD) in Alarm calls were used during the nestlingperiod and northern Arizona. food-beggingcalls during the fledgling period. For broadcasttransects, boundaries ofall tree-harvestareas METHODS weretranscribed onto 7.5-min USGS quadranglemaps. Transectswere then drawn with an east-westor north- SURVEYAREA south orientation depending on topography and ac- Areas surveyed for goshawkswere on the Kaibab cessibility,with both endsof all transectsextended 800 Plateau, Coconino County, Arizona. The Plateau is m beyond the harvest area boundaries.Distances be- boundedby escarpmentsand steepslopes that descend tween transectsand stations,as describedin Kennedy into the Grand Canyon of the Colorado River on the and Stahlecker(unpubl. data), were modified to in- south side and desert scrublandselsewhere. Forested creasethe theoreticalcoverage ofbroadcasts from 78.5 areason the NKRD, which encompassesthe northern to 90.6% of surveyedareas (Fig. 1). Modificationswere portion of the Plateau, total approximately 3000 km2 that parallel transectswere 260 m (vs. 300 m) apart;

24 GOSHAWK BROADCAST SURVEYS-Joy et al. 25

300 m ansect

B. 25 STATIONS, 1300 X 1500 m, A. 25 STATIONS, 1500 X1500 m, 90.6% COVERAGE 78.5% COVERAGE FIGURE 1. Theoretical area of coverage(circles) using (A) Kennedy and Stahlecker’s (unpubl. data) transect and station layout, vs. (B) the modified layout used in this study.

and broadcast stations were 300 m (vs. 150 m) apart per’s, Sharp-shinned, and Red-tailed hawk vocaliza- on each transect and were staggeredby 150 m on ad- tions and sightingsdetected between stationswere also jacent transects. recorded. All responsesby Accipiter spp. were imme- All broadcast stations within 800 m of active nests diately followed with a searchfor nestswithin a 200-m were deleted from the survey effort. In addition, we radius of the response.Responses by the same species did not survey stations in large, treelessopenings (di- at consecutive stations on a transect were treated as ameter > 400 m), on sheer canyon walls, or in stands one response. of Gambel’s oak (Quercus gambelii), New Mexican In 1992, broadcast surveys outside timber harvest locust (Robinia neomexicana),or pinyon pine-Rocky areas were conducted within 2 km of three goshawk Mountain juniper (L scopulorum)that did not contain nests that were active in 1991 but inactive in 1992 to large ponderosapine. search for alternate nests. An alternate nest is one of Prior to surveys, broadcast personnel were trained several nests used within a goshawk’s home range in the broadcast technique, field methods, and iden- (Reynolds and Wight 1978). In each area, transectand tification of the calls and physical characteristicsof station layout and the protocol described above were goshawks, Cooper’s (A. cooperii), Sharp-shinned (A. used. striatus), and Red-tailed (Buteo jamaicensis) hawks, and goshawkmimics (e.g., Cyanocittaspp.). Observers HAWK RESPONSES used the 7.5-min USGS maps with marked transects Chi-squaregoodness of fit tests(Ott 1984) were used and stations in the field. At each station, the surveyor to test the frequency of goshawk responsesbetween broadcastedthree times in three directions. First, the sexes, ages (adult, juvenile), nesting stages(nestling, surveyor rotated 60” right or left (determined random- fledgling)(weighted by the number of stationssurveyed ly) from the direction of travel, played the taped gos- during stage), and type of vocalizations broadcasted hawk vocalization for 10 set, and then listened and (alarm, food-begging)(weighted by number of stations searched in all directions for hawk responsesfor 30 surveyed with that vocalization) against the null hy- sec. The broadcast and observation procedureswere pothesisof equal frequenciesbetween categories.Dur- then repeated two more times after rotating 120” from ing 1992 the alarm call was inadvertently used during the previousbroadcast. Broadcasts were conductedfrom the first nine days of the fledgling stage.To determine 3 June to 12 August between 08:OOand 16:30 MDT. the effect of this, we performed simultaneous tests To avoid misidentifying broadcastsof co-workers, (Goodman 1964) of responserates to ( 1) alarm call use surveyors worked at least two transects apart. When (2 110 stations)during the nestling stage,(2) alarm call they elicited a response,the following were noted: tran- use (1058 stations) during the fledgling stage, and (3) sect, station, habitat, time, species,sex and age of the food-beggingcall use (368 1 stations) during the fledg- hawk, responsetype, bearing, and estimated initial dis- ling stage, after standardizing by number of stations tance to the respondinghawk. All goshawk, and Coo- ‘surveyed in each category.We also tested whether the 26 STUDIES IN AVIAN BIOLOGY NO. 16

successof finding a goshawk nest in follow-up nest TABLE 1. FREQUENCY OF ADULT NORTHERN searcheswas independent of the hawks’ sex and re- GOSHAWK &P~NSES BY SEX TO DIFFERENT BROAD- sponsetype. To examine trends in responseparame- CA~TVWALIZATIONSANDD~RING DIFFERENTNIZSTING ters, Fishers’ Exact test (Proc FREQ; SAS Institute STAGESINBROADWTSURVEYS ONTHE KAIBAJI PLA- 1987) was used in multi-way comparisonsof (1) re- TEAU,ARIZONA, 1991-1992 sponsetype (vocal non-approach,silent approach,vocal approach)by sex, by age, by nestingstage, and by Responses type of broadcastvocalization; and (2) sex of the re- Variable Male Female sponderby nestingstage, and by type of vocalization broadcast.We did not test for differencesin goshawk Vocalization broadcast responsefrequency among times of day. Kimmel and Alarm call 21 15 Yahner (1990) found that responserates were inde- Food-beggingcall 6 6 pendent of time of day. Among-speciescomparisons of responsesof gos- Nesting stage hawks, and Coopers,’ Sharp-shinned,and Red-tailed Nestling 15 11 hawksincluded the frequencyof responsetype, habitat Fledgling 12 10 in which the responseoccurred, and mean distance from broadcast station to respondingindividual. In multi-species comparisons, Fisher’s Exact Test (Proc FREQ; SAS Institute 1987) was used to test for differ- of survey time. One goshawk response was re- encesamong frequenciesof responsevariables, and the corded during the 1992 surveys for alternate nests. Kruskal-Wallis Test (Proc NPARl WAY; SAS Institute Six and nine active goshawk nests were located 1987) was usedto test for differencesin mean response during broadcast surveys in 199 1 and 1992, re- distancesamong species. spectively. In each year, one of these nests had SURVEYEFFORT AND ASSOCIATEDCOSTS already been found by a non-broadcast surveyor (Reynolds et al., this volume), but was found again We estimated the effort (surveyor-hours)and costs of using the broadcast survey procedure only during by a naive broadcast surveyor. Thus, one nest 1992 broadcast surveys. Our estimates included pre- was found for every 458 stations covered, 30.6 paring survey maps, training, commuting, conducting km2 surveyed, and 112.7 survey hours in 1991; surveys, daily data transcription to master 7.5min in 1992, one nest was found for every 414 sta- maps, testing and maintaining equipment, data entry, tions, 29.4 km2, and 100.3 survey hours. In ad- checking and analysis, and reporting of results. Ex- dition, the number of Coopers’ and Sharp- pensesare based on salaries,vehicle rents and mileage, shinned hawk responses during 1991 and 1992 and equipment costs and maintenance. were 14 and 8, and 15 and 6, respectively. Two Coopers’ and six Sharp-shinned hawk, and two RESULTS Coopers’ and three Sharp-shinned hawk nests were found during follow-up searches in 199 1 In 199 1 we surveyed proposed tree-harvest sites and 1992, respectively. (total area = 183.7 km2) in the Big Burro, Jack, Mean estimated hatching and fledging dates at Jolly, Burnt Saddle-Sowat, Lookout Canyon, goshawk nests were 8 June and 17 July in 199 1, Paris, and Stina areas. In 1992 we surveyed pro- and 1 June and 5 July in 1992, respectively. In posed tree-harvest sites (total area = 2 19.5 km* 199 1 broadcasting occurred during 18 days (37 + 45.2 km2 beyond harvest boundaries) in the surveyor-days) of the nestling period and 3 1 days West Lake, Holy Rock, Road Hollow, Lost Can- (86 surveyor-days) of the fledgling period. In 1992 yon, and Taters areas. A total of 6477 stations broadcasting occurred during 19 days (77 sur- were covered requiring 1579 hours of survey time. veyor-days) of the nestling period and 26 days An additional 27.8 km2 (391 stations) were cov- (117 surveyor-days) of the fledgling period. ered during surveys for alternate nests. Broadcasts of the goshawk alarm call were re- placed by broadcasts of the food-begging call on HAWK RESPONSES 22 July 199 1 and 16 July 1992, coinciding with In 199 1, 4 1 goshawk responses were detected the latest egg-hatching dates of 24 July 199 1 and during broadcast surveys (2749 broadcast sta- 15 July 1992. tions, 676 survey-hours) on the seven tree-har- There was no significant difference in response vest areas, resulting in 1.5 responses per 100 sta- frequency between adult male and female gos- tions surveyed, 0.2 responses per km2 surveyed, hawks (x2 = 0.75, df = 1, P = 0.386) during 1991 and 1 response per 16.5 hours of survey time. In and 1992 broadcast surveys. Furthermore, the 1992 surveys (3728 stations, 903 survey-hours) frequency of response by sex (in 32 cases the sex 48 goshawk responses were detected on five tree- could not be determined) was not affected by the harvest areas, resulting in 1.3 goshawk responses type of vocalization broadcasted (N = 48, Fish- per 100 stations surveyed, 0.2 responses per km2 ers’ Exact test, P = 0.741) or stage of nesting (N surveyed, and 1 response for every 18.8 hours = 48, Fishers’ Exact test, P = 1.000) (Table 1). GOSHAWK BROADCAST SURVEYS--Joy et al. 27

TABLE 2. FREQUENCYOF NORTHERN Gosn~wx RESPONSE T~PFCSBY SEX, AGE, NETTING STAGE, AND BROADCAST TYPE DURING BROAIXA~T SURVEYSFOR NORTHERN GOSHA~K.Y ON THE KA~BABPLATEAU, ARIZONA, 199 l-l 992. N = N~JMBEROF STATIONS SURVEYED

Vocal Variable non-approach Silent approach Vocal approach Sex Male 1 17 8 Female 3 3 15

Age Adult 4 12 11 Juvenile 9 0 3 Nesting stage Nestling (N = 2 129) 9 15 17 Fledgling (N = 4739) 18 13 14 Vocalization broadcasted Alarm call (N = 3 168) 12 23 23 Food-begging call (N = 3700) 15 5 8

Response type, however, differed between the or vocal approach, whereas food-begging calls sexes (N = 47, Fishers’ Exact test, P = 0.001) more often elicited vocalization with no ap- (Table 2); males often approached silently, proach. More nests were found during follow-up whereas females most often approached and vo- searches to vocal approaches (N = 3 1 nests) than calized. More active nests were found in follow- to vocal non-approaches (N = 27 nests) or silent up searches to responses by females (N = 8 nests approaches (N = 28 nests) alone (N = 86, Fishers’ of 21 responses) than males (N = 3 nests of 27 Exact test, P = 0.029). responses)(N = 48, Fishers’ Exact test, P = 0.040). The overall chi-square among the three cate- During the fledgling stage, adults responded to gories of call-use by nesting stage (alarm call dur- broadcasts more often than juvenile hawks (x2 ing the nestling stage, alarm call during the fledg- = 7.05, df = 1, P = 0.008; N = 29 and 12 re- ling stage, and food-begging call during the sponses, respectively). Also, adults were more fledgling stage) was significant (x2 = 14.90, df = likely to approach the broadcaster (either vocal- 2, P = 0.013). Response rates were significantly izing or silent) than juveniles (N = 93, Fishers’ higher using the alarm call during the nestling Exact test, P < 0.001) (Table 2). stage than using the food-begging call during the During both years, more stations were sur- fledgling stage. However, goshawk response rates veyed during the fledgling (4739 stations) than using the alarm call during the fledgling period the nestling (2 129 stations) stage. After adjusting were not significantly different from using the for these differences, goshawks responded more food-begging call during the fledgling period, nor often during the nestling stage (42 responses per did the former differ from using the alarm call 2129 stations) than the fledgling stage (48 re- during the nestling period. sponses per 4739 stations) (x2 = 10.33, df = 1, Overall, the frequency of response types among P = 0.00 l), whereas response type was indepen- the three species of Accipiter did not differ (N = dent of nesting stage (N = 86, Fishers’ Exact test, 123, Fishers’ Exact test, P = 0.208), nor did they P = 0.202) (Table 2). differ between any Accipiter and Red-tailed The number of stations surveyed with the alarm Hawks (Fishers’ Exact test, P > 0.016, Bonfer- and food-begging calls also varied (alarm call = roni significance level, for all pairwise tests). 3 168 stations, food-begging call = 3700 stations). However, when responses were partitioned by After adjustment, goshawks responded more of- habitat (ponderosa pine, mixed-conifer forests), ten to the alarm (N = 59 responses/effort) than the frequency of responses among all species dif- food-begging (N = 3 1 responses/effort) call (x2 = fered (N = 2 10, Fishers’ Exact test, P = 0.014) 13.69, df = 1, P < 0.001). Response type (4 of (Table 3). Goshawk and Coopers’ and Red-tailed 90 responses were unclassified) was not inde- hawks responded more often in ponderosa pine pendent of the type of call broadcasted (N = 86, forest, whereas Sharp-shinned Hawks responded Fishers’ Exact test, P = 0.008) (Table 2). Alarm more often in mixed-conifer forests. Mean re- calls more commonly elicited a silent approach sponse distance was not equal among species (P 28 STUDIES IN AVIAN BIOLOGY NO. 16

TABLE 3. FREQUENCYOF NORTHERNGOSHAWK, Vehicle and equipment costs were $1.90 per sta- COOPER'S HAWK, SHARP-SHINNED HAWK, AND tion and $26.70 per km2 ($69.30/m?) surveyed. RED-TAILEDHAWK RESPONSESBY HABITAT DUIZING BROADCASTSURVEYS ON THE KAIBAB PLATEAU, DISCUSSION ARIZONA,1991-1992 HAWK RESPONSES

Species Response rates of goshawks to broadcasts did not differ between the sexes, but male and female Sharp- Red- Northern Coopers’ shinned tailed behaviors differed when responding. As mea- Habitat Goshawk Hawk Hawk Hawk sured by their intensity of vocalization, females Ponderosapine 50 17 5 62 responded more aggressively than males. Males Mixed-conifer 38 7 9 22 appeared to be more inquisitive than aggressive in their initial response, often approaching and retreating silently. The probability of finding a nest during follow-up searches to female re- < 0.001); Red-tailed Hawks responded at dis- sponses was greater than for male responses. Fe- tances greater than any of the Accipiter (P < males are more likely to be in the vicinity of their 0.016, Bonferroni significance level, in all pair- nests during early stages of brood-rearing where- wise tests) (Table 4). Response distances did not as males are more likely to be away from nests differ among the Accipiter species (P = 0.492). foraging. However, because of potential bias in the misidentification of sex, and because males SURVEY EWORTS AND ASSOCIATEDCOSTS will respond from their nests, we recommend In 194 surveyor-days during 1992, observers that equal effort be given to follow-up searches surveyed 3728 stations (265 km2 of transects) for to responses by either sex. a mean survey effort of 19.3 (SD = 6.8) stations Contrary to Kimmel and Yahners’ (1990) re- per day. Of the 4832 established stations (343 port that no difference occurred in response rates km2), 1104 stations (78.4 km*) were not surveyed between nesting stages, we found that goshawk because they occurred in non-forested habitat or responses rates were higher during the nestling were within 800 m of an active goshawk nest. than fledgling stage. No difference in response Eleven observers spent 0.5-10.0 h (x = 4.7 h, rates to the alarm and food-begging calls during SD = 1.7) per day surveying for a total survey the fledgling stage was detected during our sur- time of 903.1 h. Mean time required to broad- veys. Kennedy and Stahlecker (1993), however, cast, observe, and travel to the next station was detected fewer responses using the alarm call vs. 14.5 min per station. A mean of 8.4 h per day the food-begging call during this period. (SD = 1.7, total time = 1627 h) per observer was During the fledgling stage, adult response type required to broadcast, commute to and from sur- (silent and vocal approaches to the observer) dif- vey sites, update master survey maps, and test fered from juvenile response type (vocal non- and maintain equipment. An additional 184 sur- approaches). This difference may reflect the in- veyor-hours were required for training personnel ability of fledglings to fly well and a reluctance and 252 h were required on pre- and post-season to leave the nest area. Juveniles vocally respond- tasks such as preparing survey maps, selecting ed with both alarm and food-begging calls. personnel, data entry and verification, analysis, Response type to the calls broadcasted (alarm, and report writing. Total personnel costs to com- food-begging), when both nesting stages were plete broadcast surveys of the five tree-harvest combined, differed. More approaches, with and areas surveyed in 1992 ranged from $4.15-5.80 without vocalization, were tallied in response to per station and $58.50-81.85 per km2($151.40- the alarm call whereas more vocalizations with- 21 1.95/miz) depending on salary (Appendix). out approach were tallied to the food-begging

TABLE 4. MEANINITIAL-RESPONSES DISTANCE, STANDARD DEVIATION, AND NUMBERS OF NORTHERN G~SHAWK, COOPERSHAWK,’ SHARP-SHINNED HAWK, AND RED-TAILEDHAWKRESPONSESTOBROADCMT Suavsvs ONTHE KAIBAB PLATEAU, ARIZONA,1991-1992

SpeCkS Northern C;=CW& Sharp-shinned Red-tailed Variable Goshawk Hawk Hawk Mean responsedistance (m) 95.4 75.8 77.0 148.1 SD 87.5 71.1 73.5 121.3 No. of observations 90 26 14 89 GOSHAWK BROADCAST SURVEYS--Joy et al. 29 call. This pattern reflects the tendency of adults D. Z. Rubinoff, D. S. Strait, and D. J. Worthington for to respond to the alarm call and juveniles to the their hard work in the field. Thanks to the Arizona food-begging call. A high number of vocal re- Game and Fish Department for providing J. T. Dris- ~011,B. S. Heslin, T. L. Kelley, T. L. Moore, T. J. Rice, sponses without approach were also tallied by and L. E. Schultz to help with surveys. J. G. Dudley Kennedy and Stahlecker (1993) during the fledg- provided valuable input in modifying the Kennedy and ling stage. Stahleckerbroadcast design. For statisticaladvice, we Forests surveyed during 1991 and 1992 were thank R. M. King and J. E. Janssen.We also thank P. primarily ponderosa pine and mixed-conifer. A H. Bloom, D. Call, and L. G. Eskew for their valuable greater proportion of goshawk, and Coopers’ and reviews of the manuscript. Funding, technical assis- Red-tailed hawk responses to goshawks vocali- tance, and housing were graciously provided by the zations were obtained in ponderosa pine than in North Kaibab Ranger District, USDA Forest Service. mixed-conifer forests. For Sharp-shinned Hawks, LITERATURE CITED a greater proportion of responses were obtained CROCKER-BEDFORD,D. C. 1990. Goshawk reproduc- in mixed-conifer forest. Mixed-conifer forests are tion and forest management. Wildl. Sot. Bull. 18: often more structurally and floristically diverse 262-269. than ponderosa pine forests and probably pro- GOODMAN, L. 1964. Simultaneous confidence inter- vide greater hiding cover for Sharp-shinned Hawk vals for contrasts among multinomial populations. nests and higher densities of their prey. Initial Ann. Math. Stat. 35:716-725. response distance to broadcasts was greater for KENNEDY. P. L.. AND D. W. STAHLECKER. 1993. Re- Red-tailed Hawks than any of the Accipiter spp. sponsiveness bf nesting Northern Goshawks to taped broadcasts of three conspecific calls. J. Wildl. Man- Red-tailed Hawks soar more often than the Ac- age. 571249-257. cipiter and are easier to detect at greater dis- KIMMEL, J. T., AND R. H. YAHNER. 1990. Response tances. of Northern Goshawks to taped conspecific and Great We think that the Kennedy and Stahlecker Homed Owl calls. J. Raptor Res. 24:107-l 12. (1993) protocol (with modifications described MOSHER, J. A., M. R. FULLER, AND M. KOPENY. 1990. herein) is an effective means of locating goshawk, Surveying woodland raptors by broadcast of con- and Coopers’ and Sharp-shinned hawk nests. specific vocalizations. J. Field Omith. 6 1:45346 1. Blind tests of the efficacy of the broadcast-survey OTT, L. 1984. An introduction to statistical methods method need to be conducted to determine the and data analysis. Duxbury Press, Boston, MA. RASMUSSEN,D. I. 1941. Biotic communities of Kai- proportion of nests missed and the number of bab Plateau, Arizona. Ecol. Monogr. 3:1-275. survey visits required to achieve an acceptable REYNOLDS, R. T. 1983. Management of western co- level of detection. Until such tests are performed, niferous forest habitat for nesting accipiter hawks. guidelines for surveying proposed tree-harvest USDA Forest Service, Gen. Tech. Rep. RM- 107, Ft. areas should require two or more visits in sep- Collins, CO. arate years to increase the probability of goshawk REYNOLDS, R. T. 1989. Accipiters. Pp. 92-101 in B. detection. Tests of the efficacy of using goshawk G. Pendleton, C. E. Ruibal, D. L. Krahe, K. Steen- vocalizations versus those of congeners to find hof, M. N. Kockert, and M. L. Le Franc, Jr. (eds.), Coopers,’ Sharp-shinned, and Red-tailed hawk Proceedings ofthe Western raptor management sym- posium and workshop. Natl. Wildl. Fed. Sci. Tech. nests should be pursued. Ser. No. 12. SURVEY EFFORT AND ASSOCIATEDCOSTS REYNOLDS, R. T., R. T. GRAHAM, H. M. REISER, R. L. BASSETT, P. L. KENNEDY, D. A. BOYCE, JR., G. Estimates of cost/effort of locating goshawk GOODWIN, R. SMITH, AND E. L. FISHER. 1992. Man- nests using methods similar to Kennedy and agement recommendations for the Northern Gos- Stahlecker (1993) were not available for com- hawk in the Southwestern United States. USDA For- parison with our estimates. Costs associated with est Service, Gen. Tech. Rep. RM-217, Ft. Collins, surveying for goshawks (personnel, vehicle, and co. equipment) vary depending on survey effort and REYNOLDS, R. T. AND H. M. WIGHT. 1978. Distri- bution, density, and productivity of Accipiter hawks size of the survey area. If finding all goshawk breeding in Oregon. Wilson Bull. 90: 182-196. nests is the objective of surveys, effort should not SAS INSTITUTE. 1987. SAS/STAT Guide for Personal be compromised in order to cover larger areas. Computers, Version 6. SAS Institute Inc., Cary, NC. ACKNOWLEDGMENTS We thank S. T. Bedard, K. L. Boden, P. C. Clark, S. E. Dudley, J. G. Dudley, M. W. Gavin, K. J. Kalin,